[Pacemaker] pacemaker processes RSS growth

Vladislav Bogdanov bubble at hoster-ok.com
Fri Sep 7 02:25:54 EDT 2012


06.09.2012 12:58, Vladislav Bogdanov wrote:
...
> lrmd seems not to clean up gio channels properly:

I prefer to call g_io_channel_unref() right after g_io_add_watch_full()
instead of doing so when deleting descriptor (g_source_remove() is
enough there). So channel will be automatically freed after watch fd is
removed from poll with g_source_remove(). If I need to replace watch
flags, I call g_io_channel_ref()/g_io_channel_unref() around
g_source_remove()/g_io_add_watch_full() pair. This would require some
care and testing though. I have spent two days on this when writing my
first non-blocking server based on glib and gio. This code now works
like a charm with approach outlined above.

> 
> ==1734== 8,946 (8,520 direct, 426 indirect) bytes in 71 blocks are
> definitely lost in loss record 147 of 152
> ==1734==    at 0x4C26FDE: malloc (vg_replace_malloc.c:236)
> ==1734==    by 0x71997D2: g_malloc (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x71C67F4: g_io_channel_unix_new (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x4E52470: mainloop_add_fd (mainloop.c:660)
> ==1734==    by 0x5067870: services_os_action_execute (services_linux.c:456)
> ==1734==    by 0x403AA6: lrmd_rsc_dispatch (lrmd.c:696)
> ==1734==    by 0x4E513C2: crm_trigger_dispatch (mainloop.c:105)
> ==1734==    by 0x7190F0D: g_main_context_dispatch (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194937: ??? (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194D54: g_main_loop_run (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x402427: main (main.c:302)
> ==1734==
> ==1734== 8,946 (8,520 direct, 426 indirect) bytes in 71 blocks are
> definitely lost in loss record 148 of 152
> ==1734==    at 0x4C26FDE: malloc (vg_replace_malloc.c:236)
> ==1734==    by 0x71997D2: g_malloc (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x71C67F4: g_io_channel_unix_new (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x4E52470: mainloop_add_fd (mainloop.c:660)
> ==1734==    by 0x50678AE: services_os_action_execute (services_linux.c:465)
> ==1734==    by 0x403AA6: lrmd_rsc_dispatch (lrmd.c:696)
> ==1734==    by 0x4E513C2: crm_trigger_dispatch (mainloop.c:105)
> ==1734==    by 0x7190F0D: g_main_context_dispatch (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194937: ??? (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194D54: g_main_loop_run (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x402427: main (main.c:302)
> ==1734==
> ==1734== 65,394 (62,280 direct, 3,114 indirect) bytes in 519 blocks are
> definitely lost in loss record 151 of 152
> ==1734==    at 0x4C26FDE: malloc (vg_replace_malloc.c:236)
> ==1734==    by 0x71997D2: g_malloc (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x71C67F4: g_io_channel_unix_new (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x4E52470: mainloop_add_fd (mainloop.c:660)
> ==1734==    by 0x5067870: services_os_action_execute (services_linux.c:456)
> ==1734==    by 0x50676B4: recurring_action_timer (services_linux.c:212)
> ==1734==    by 0x719161A: ??? (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7190F0D: g_main_context_dispatch (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194937: ??? (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194D54: g_main_loop_run (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x402427: main (main.c:302)
> ==1734==
> ==1734== 65,394 (62,280 direct, 3,114 indirect) bytes in 519 blocks are
> definitely lost in loss record 152 of 152
> ==1734==    at 0x4C26FDE: malloc (vg_replace_malloc.c:236)
> ==1734==    by 0x71997D2: g_malloc (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x71C67F4: g_io_channel_unix_new (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x4E52470: mainloop_add_fd (mainloop.c:660)
> ==1734==    by 0x50678AE: services_os_action_execute (services_linux.c:465)
> ==1734==    by 0x50676B4: recurring_action_timer (services_linux.c:212)
> ==1734==    by 0x719161A: ??? (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7190F0D: g_main_context_dispatch (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194937: ??? (in /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x7194D54: g_main_loop_run (in
> /lib64/libglib-2.0.so.0.2200.5)
> ==1734==    by 0x402427: main (main.c:302)





More information about the Pacemaker mailing list